Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Belgique → Developing pedotransfer function to predict the soil water retention curve of Kenyan soils

Universiteit Gent (2010)

Developing pedotransfer function to predict the soil water retention curve of Kenyan soils

Nyambura Geofrey Waweru

Titre : Developing pedotransfer function to predict the soil water retention curve of Kenyan soils

Auteur : Nyambura Geofrey Waweru

Université de soutenance : Universiteit Gent

Grade : Master of Science (MS) Physical Land Resources 2010

Résumé
The soil water retention curve (SWRC) is very important in soil and water management. It has many applications in agricultural, ecological, and environmental studies. However, despite these vital roles, information on SWRC is not readily available because the direct measurement of the soil hydraulic properties is laborious, timeconsuming and expensive. Therefore, indirect estimation of SWRC and bulk density is the most viable option to generate this information. This study was carried out to evaluate some published PTFs and develop PTFs for predicting SWRC and bulk density of Kenyan soils. 100 undisturbed soil samples were taken from top and sub horizons in 25 soil profiles in Kenya. Validation indices such as mean error, standard deviation prediction error, root mean square error, and Pearson correlations were used to quantify the PTFs’ predictability. Land use, bulk density, organic matter, CEC and plastic limit were found to be potential predictors. The evaluated PTFs performed poorland cannot be applied to estimate soil water content. The study showed poor correlation between soil granulometric fraction with SWRC and bulk density. This was attributed to the different soil types with contrasting properties. Efforts were done to separate the soils into different groups to improve the correlation but the number of soil samples per each soil type was too few to be statistically acceptable. The PTF developed for the van Genuchten parameters to predict SWRC using these predictor variables showed r value of 0.94. A model to predict bulk density was developed with r value of 0.76. However, the PTFs need to be validated with independent data set

Source : Pedon 22 - Physical Land Resources - Universiteit Gent

Page publiée le 1er janvier 2016, mise à jour le 25 janvier 2018